Conyza sumatrensis Resistant to Paraquat, Glyphosate and Chlorimuron: Confirmation and Monitoring the First Case of Multiple Resistance in Paraguay

Conyza sumatrensis was reported to be associated with 20 cases of herbicide resistance worldwide, with a recent report of multiple drug resistance to paraquat, glyphosate, and chlorimuron in Brazil. In Paraguay, there were no reports of cases of resistance for this species; however, in 2017, researchers began identifying biotypes with resistance to paraquat, glyphosate, and chlorimuron, which is the focus of the present study. The goal of this study was to investigate the case of multiple resistance of C. sumatrensis to paraquat, glyphosate, and chlorimuron and to monitor the resistant biotypes in the departments of Canindeyú and Alto Paraná. Seeds were collected from sites where plants survived after herbicide application in the 2017/18 and 2018/19 seasons. After screening, biotypes were selected for the construction of dose–response curves. A resistance factor (RF) of 6.79 was observed for 50% control (C50) and 3.92 for 50% growth reduction (GR50) for the application of paraquat. An RF of 12.32 was found for C50 and 4.15 for GR50 for the application of glyphosate. For the application of chlorimuron, an RF of 11.32 was found for C50 and 10.96 for GR50. This confirms the multiple resistance of the C. sumatrensis biotype to paraquat, glyphosate, and chlorimuron. Population monitoring indicated the presence of C. sumatrensis with multiple resistance in departments of Canindeyú and Alto Paraná, Paraguay.


Seed Collection
Seeds were collected in sites where C. sumatrensis plants survived after herbicide burndown application in pre-sowing in the 2017/18 and 2018/19 growing seasons, in 33 agricultural areas located in the departments of Canindeyú and Alto Paraná, Paraguay. The geographical coordinates, biotype identification, and infested crops are listed in Table 1. Among these sites, there are two locations with possibly susceptible plants that served as a comparison control.
The sampling sites were chosen according to reports of control failures as sites with possible cases of resistance. Our seed collection followed the methodology proposed by Burgos et al. [15]. For each site, seeds were collected from 5-10 plants, with the same characteristics, pooled into a single sample per site (with at least 1000 physiologically mature seeds per sample).

Screening
In a greenhouse, with daily irrigation, in the municipality of Katueté, Canindeyú, Paraguay (24 • 09 28.7" S, 54 • 52 10.4" W), about 100 seeds were sown in a plastic tray filled with substrate potting mix, for each sampling site, from October to November 2018. After germination, seedlings were transplanted into 800 mL plastic pots and filled with substrate potting mix with one seedling per pot. A completely randomized design with eight replications was used for each herbicide applied. Six herbicides, at the average recommended dose for the control of C. sumatrensis at the stage of 6-8 true leaves, were applied to plants, in addition to the control (no application) ( Table 2), for each sampling site. The application took place at the stage of 6-8 true leaves, using a series 110.02 (TeeJet Technologies, Wheaton, IL, USA) CO 2 backpack sprayer pressurized at a constant pressure of 2 kgf cm −2 , with a bar with four fan nozzles, positioned at 50 cm from the target and at a speed of 1 m s −1 , providing a total spray volume of 200 L ha −1 .
Plant control was evaluated at 28 days after application (DAA), and visual scores were assigned to each experimental unit, where 0 represents no damage and 100% indicates total plant death [16]. The results were presented descriptively. After screening, plants from certain populations were selected to be grown alone to generate F 1 seeds, which were used for the construction of dose-response curves. The generation of F 1 is important to attest to the inheritance of the resistance character of populations.

Dose-Response Curves
The same screening procedures were followed for sowing, seedling transplantation, and herbicide application, at the same location. The biotype whose F 1 seeds were collected and investigated for resistance came from sampling site 27 (24 • 03'34"S 55 • 00'20"W), and the susceptible biotype, also from the F 1 generation, came from site 33 (24 • 08'58"S 54 • 51'24"W). The herbicides applied were paraquat (0, 50, 100, 200, 400, 800, 1600, and 3200 g active ingredient (ai) ha The plant control was evaluated at 28 DAA; visual scores were assigned to each experimental unit, where 0 indicates no damage and 100% indicates total plant death [16]. Dry mass evaluation was carried out at 28 DAA of the herbicides. Plants were cut at the ground level, placed in paper bags, dried in an oven at 70 • C for four days (to reach constant mass), and then measured.

Statistical Analysis
After screening for the generation of heritability (F 1 ), selection of biotypes, and realization of dose-response curves, the data of the evaluations from 28 DAA were subjected to analysis of variance and regression (p ≤ 0.05) and adjusted for the nonlinear logistic regression model proposed by Streibig [17]: where y is the response variable (percentage control or shoot dry mass); x is the herbicide dose (g ha −1 ); and a, b, and c are the estimated parameters of the equation, so that a is the amplitude between the maximum and the minimum point of the variable, b is the dose that provides 50% response, and c is the slope of the curve around b. The non-linear logistic model provides an estimate of parameter C 50 (50% control) or GR 50 (50% growth reduction). Thus, we opted for mathematical calculation using the inverse equation of Streibig [17], allowing the calculation of C 50 , as proposed by Souza et al. [18]. The models used to obtain C 50 were the same as those used by Takano et al. [19], Takano et al. [20], and Albrecht et al. [14].
Based on the values of C 50 and GR 50 , we calculated the resistance factor (RF = C 50 or GR 50 of the resistant biotype/C 50 or GR 50 of the susceptible biotype). The resistance factor expresses the number of times that the dose required to control 50% resistant biotypes is greater than the dose controlling 50% susceptible biotypes [15,21].
Points with ≤50% control were plotted in red, from 51% to 85% in yellow, ≥86% in green, highlighting site 27 (R)-resistant to paraquat, glyphosate, and chlorimuron-and site 33 (S)susceptible to herbicides. The proximity of collection points of the resistant and susceptible biotypes is presented in Figure 1.  According to the results of the screening, the biotype from site 27 was selected to investigate the possible case of resistance to herbicides. An RF of 6.79 was observed for C 50 (Figure 2A) and 3.92 for GR 50 (Figure 2B), for the application of paraquat. The ineffectiveness in controlling C. sumatrensis under the application of glyphosate was also verified; for C 50 and GR 50 , RF was 12.32 ( Figure 2C) and 4.15 ( Figure 2D), respectively. For chlorimuron, RF was 11.32 for C 50 ( Figure 2E) and 10.96 for GR 50 ( Figure 2F). This confirmed the triple resistance of the C. sumatrensis biotype (site 27) to the herbicides paraquat, glyphosate, and chlorimuron (Table 4). possible case of resistance to herbicides. An RF of 6.79 was observed for C50 ( Figure 2A) and 3.92 for GR50 ( Figure 2B), for the application of paraquat. The ineffectiveness in controlling C. sumatrensis under the application of glyphosate was also verified; for C50 and GR50, RF was 12.32 ( Figure 2C) and 4.15 ( Figure 2D), respectively. For chlorimuron, RF was 11.32 for C50 ( Figure 2E) and 10.96 for GR50 ( Figure 2F). This confirmed the triple resistance of the C. sumatrensis biotype (site 27) to the herbicides paraquat, glyphosate, and chlorimuron (Table 4).

Discussion
The low efficiency of paraquat, glyphosate, and chlorimuron was observed in most areas where C. sumatrensis seeds were collected. Control of ≥86% was observed in only two sites, for the three herbicides simultaneously. The identification of biotypes resistant to the three herbicides demonstrated the low effectiveness of these herbicides in controlling C. sumatrensis in a large area. The low effectiveness Agriculture 2020, 10, 582 8 of 11 of these herbicides against C. sumatrensis was been reported in Brazil, including in states bordering Paraguay (Paraná and Mato Grosso do Sul). This low efficacy was confirmed by the cases of simple and multiple resistance to paraquat, glyphosate, and chlorimuron [14,22,23]. Albrecht et al. [14] showed multiple resistance to paraquat, glyphosate, and chlorimuron with RF for the C 50 of 7.43, 3.58, and 14.35 and for the GR 50 of 2.65, 2.79, and 11.31, respectively. In the present study, we observed RF for the C 50 of 6.79, 12.32, and 11.32 and for the GR 50 of 3.92, 4.15, and 10.96, respectively, for paraquat, glyphosate, and chlorimuron-that is, with RF close to paraquat and chlorimuron in the comparison between these biotypes. A higher RF was found for glyphosate in the biotype identified in Paraguay in this study.
In contrast, the herbicides saflufenacil and glufosinate were effective in controlling C. sumatrensis in all sampling sites, and the herbicide 2,4-D also showed good control; however, 2,4-D and other synthetic auxins are the subject of other specific studies due to the rapid necrosis, as studied in Brazil [24]. This reinforces the need to use different herbicides to control weeds, focusing not only on management, but also on preventing the selection of new resistant biotypes. Other studies demonstrated the effectiveness of these herbicides in the control of species of the genus Conyza [25][26][27][28][29]-in most situations, in combination with other herbicides, including products with confirmed resistance.
The combination and rotation of herbicides with different mechanisms of action are reinforced by several studies as essential in preventing the selection of new cases, in the effective management of already resistant cases, and in expanding the spectrum of action of the herbicidal treatment [30][31][32]. In addition, non-chemical measures, such as cover crops, should be highlighted. For example, vetch and barley crop residues were effective in suppressing C. canadensis [33], and black oat and wheat in suppressing C. bonariensis [34]. The importance of monitoring the populations of resistant weeds is therefore emphasized, which allows for the identification of the evolution and dispersion of cases of resistance, which consequently provides subsidies for decision-making for the effective management of weeds [35,36]. This study highlights the importance of and identifies the levels of effectiveness of herbicides in the region where the biotype was recorded.
The monitoring weed resistance cases is, therefore, an essential practice to understand, identify, and quantify the frequency of these plants in advance [37]. Thus, studies on resistance monitoring lead to increased research and, consequently, new techniques for the control of problematic plants, such as the use of pre-emergent herbicides to decrease the selection pressure [38][39][40].
In Paraguay, only four cases of herbicide-resistant weed biotypes have been officially reported, including the present study. In addition to this, Euphorbia heterophylla was found to be resistant to imazethapyr (an ALS inhibitor), and Digitaria insularis and Bidens subalternans were resistant to glyphosate [1]. This reinforces the importance of the present study, not only by identifying the first case of multiple resistance in the country, but also for monitoring the population of C. sumatrensis and investigating the effectiveness of herbicides. This provides important information for the management of this weed and for prevention of the selection of new resistant biotypes.
This population of C. sumatrensis meets all the criteria set to confirm a new case of resistance to paraquat, glyphosate, and chlorimuron, according to the criteria for confirming a new case of weed resistance to a herbicide of the Herbicide Resistance Action Committee (HRAC) [41]. These criteria include the definition of weed resistance; confirmation of the results obtained by scientifically based protocols; characterization of the heritability of weed resistance to the herbicide; demonstration of the practical impact in the field of weed resistance to the herbicide; and botanical identification of the weed species under analysis and not as a result of deliberate/artificial selection. This case was reported to the International Herbicide-Resistant Weed Database and is already registered [1].

Conclusions
Our results confirmed the multiple drug resistance of C. sumatrensis to the herbicides paraquat (a photosystem I inhibitor), glyphosate (an EPSPs inhibitor), and chlorimuron (an ALS inhibitor) as all the criteria set to prove new cases of resistance of weeds were met, thus scientifically demonstrating the first case of a weed with multiple resistance to herbicides in Paraguay.
Population monitoring indicated the presence of C. sumatrensis plants with triple multiple resistance in the departments of Canindeyú and Alto Paraná, Paraguay, in most of the sampled sites. Further monitoring research on this weed species is ongoing in Paraguay, also covering the suspected resistance to 2,4-D and for other weed species, due to the scarcity of results in this country. Studies are underway with the objective of characterizing effective and sustainable alternatives for the control of this weed.